Amazon: Abandoning the monopoly strategy of a single model, multi-model ecology becomes the core of AI layout

Amazon's second-quarter financial report has attracted industry attention. The report shows that its core business, Amazon Web Services (AWS), achieved a strong revenue growth of 37%. Behind this impressive performance reflects the company's strategic shift in the field of artificial intelligence (AI) — abandoning the pursuit of a monopoly on a single top model and instead building an ecosystem with multiple models coexisting.

Recently, discussions about Amazon's progress in independently developing AI models facing obstacles have frequently appeared in the news. However, Amazon CEO Andy Jassy emphasized during the analyst conference call on the 30th that this situation does not pose a substantial threat to the company. Jassy pointed out that the future AI industry will not be dominated by a single model, and customers are more inclined to leverage the advantages of multiple leading models in practical applications. This trend has directly driven the explosive growth of Amazon's foundational model platform, Bedrock.

"Amazon and AWS can still achieve tremendous commercial success even without being at the forefront of the industry's models," Jassy stated clearly, expressing the company's commitment to making Bedrock the world's largest inference engine. He reiterated that for companies developing critical AI applications, the ability to flexibly access all currently available models is crucial, as different models will alternate in leading positions across various fields.

This strategy marks Amazon's deviation from the industry norm of competing for the "strongest model" since the explosion of generative AI trends in 2022. As market competition deepens, the industry's focus is gradually shifting from merely pursuing model performance to the cost-effectiveness ratio of model operation (i.e., "the level of intelligence output per unit cost"), which provides an opportunity for Amazon.

In the process of reshaping its AI strategy, Amazon has not completely abandoned independent research and development. It is understood that the company is gradually reducing its investment in the original Nova series models and concentrating resources on advancing new cutting-edge model development. Jassy predicts that within the next few years, at least half a dozen top models with comparable performance will emerge in the market. He emphasized, "These models will all connect to the Bedrock platform, and there will certainly be a place for Amazon's independently developed models."

Jassy further pointed out that mastering its own leading models not only enhances control over the costs of internal consumer applications and AWS customers but also allows for more efficient responses to customer needs without relying on external suppliers. For example, its smart voice assistant Alexa has begun to reduce its dependence on high-cost third-party models and is increasingly adopting internal AI models to optimize its cost structure.

Overall, Amazon's current AI strategy has clearly focused on building an open ecosystem based on the Bedrock platform, establishing a competitive advantage by providing a wide range of model choices, rather than engaging in a head-to-head battle with competitors on single model metrics